Automated Detection and Diameter Estimation for Mouse Mesenteric Artery Using Semantic Segmentation.

Area of Science:

• Cardiovascular Research
• Medical Imaging Analysis
• Machine Learning in Biology

Background:

• Pressurized myography is crucial for assessing small artery structure and function.
• Current methods require significant technical skill for sample preparation and artery selection.
• Automating vessel identification and sizing can overcome these limitations.

Purpose of the Study:

• To develop an automated system for artery/vein differentiation and size measurement using machine learning.
• To streamline the experimental workflow for pressurized myography.
• To improve the accuracy and efficiency of small vessel analysis.

Main Methods:

• Trained a U-Net based machine learning model on 654 mouse mesenteric artery images.
• Validated the model using Intersection-over-Union and Dice coefficient metrics.
• Correlated automated measurements with manual measurements and assessed vessel size changes during pressurized myography.

Main Results:

• The model achieved high accuracy in vessel segmentation (Dice coefficient: 0.881 ± 0.016).
• Automated vessel and lumen size measurements showed strong linear correlation with manual measurements (R = 0.722 and R = 0.908, respectively).
• Developed models accurately predicted vessels meeting specific size criteria for myography.

Conclusions:

• The U-Net based image analysis method effectively automates artery/vein differentiation and size measurement.
• This approach significantly streamlines the experimental process for pressurized myography.
• The system enhances the feasibility and efficiency of small vessel research.